Description of the methodological approach

  1. In this first iteration of the living systematic review we searched for randomised controlled trials that compared pro-dopaminergic interventions to placebo in adults with unipolar depression (i.e. any standardised measure, above-threshold symptoms on any standardised measure, or a clinical diagnosis based on any operationalised criteria).

  2. Eight databases were searched from inception up to the 4th of September, 2023: Web of Science Core Collection, BIOSIS Citation Index (Web of Science), CENTRAL, EMBASE (OVID), MEDLINE (OVID), PsycINFO (OVID), and SCOPUS (see appendix for the full search strategies). Database search results were imported into EPPI-Reviewer and de-duplicated prior to screening. All steps related to record screening and data extraction were completed in EPPI-Reviewer.

  3. Titles and abstracts of the identified records were screened by at least two reviewers (CF, EB, JK, JP); conflicts were resolved through discussion within the screening team and consultation with senior reviewers (AC, EGO). We retrieved the full-texts and any supporting documents for all records that were not excluded at the title and abstract screening stage. The full-text screening was conducted by at least two reviewers (CF, CA, EB, JK, JP) and conflicts were resolved through discussion between the two reviewers and involvement of a third reviewer (EO, AC). Additional information on the full eligibility criteria can be found in the pre-published protocol.

  4. Relevant data was extracted using EPPI-Reviewer by at least two reviewers (CF, CA, EB, JK).

  5. As specified in our protocol, we focused on:

For anhedonia and anxiety symptom severity, we extracted outcome data reported at 8 weeks post-treatment or manipulation. If information at 8 weeks was not available, we considered eligible data ranging between 4 and 12 weeks (with preference to the time point closest to 8 weeks, and if equidistant the longer outcome). For acceptability, tolerability, safety and specific-adverse events, we extracted outcome data reported at the end of the studies.

  1. When extracting continuous outcomes we extracted mean and standard deviation to two decimal places. Where standard error was reported instead, we converted the value to standard deviation. Baseline and endpoint values were extracted, where only change in score and baseline or endpoint was reported, the missing value was calculated by adding or subtracting the change score to the baseline or endpoint.

  2. When extracting dichotomous outcomes we extracted natural numbers and where only percentages of participant groups were reported, a value was calculated and rounded up to the nearest natural number. Adverse events were extracted using the exact terms they were reported in the included studies.

  3. We assessed risk of bias with the RoB2 tool (Sterne et al. 2019). All outcomes for all included studies were assessed as per [CITE] by at least two reviewers (JK, CF, CA, AH) and conflicts were were resolved by discussion between the reviewers or in consultation with senior reviewers (AC, EGO). To evaluate biases due to missing evidence and biases across studies the ROB-ME tool was used with the same double screening and conflict resolution process as described above.

  4. Effect sizes were calculated as standardised mean differences (SMDs) for continuous outcomes (anhedonia and anxiety symptom severity) and odds ratios (ORs) for dichotomous outcomes (acceptability, tolerability, and specific-adverse events). We calculated the 95% confidence interval (CI) around the pooled effect size for each meta-analysis.

  5. The meta-analyses were conducted using a random effects model with the inverse variance method, using the restricted maximum-likelihood estimator for tau^2 and the Q-Profile method for the confidence interval of tau^2. Confidence intervals were adjusted using the Hartung-Knapp method. Prediction intervals were calculated to better report the effect of heterogeneity on the overall pooled effect.

  6. We conducted a series of sensitivity analyses on the primary outcome. We aggregated IPD on randomised controlled trials on antidepressants in people with depression, performing a series of random effects network meta-analyses to compare the effects of pro-dopaminergic and non-pro-dopaminergic pharmacological interventions on anhedonia.

  7. Subgroup analysis and meta-regressions were also conducted for the following variables: mean age of participants, mean anhedonia baseline score, mean anxiety baseline score, sex (proportion of female participants), mean reward baseline score, and planned treatment duration. This was done using a mixed-effects model with the estimation of the between-study heterogeneity tau2 based on the REML method. Meta-regressions were only conducted for outcomes where k > 10.

  8. Summary of evidence tables were constructed for all outcomes including a summary of the meta-analytic result, biases within-study, across-study, and due to indirectness.

Please see the final manuscript, extended data, and protocol for more details.

A list of abbreviations can be found towards the end of the document.

Results

Below are the results of the systematic review and meta-analysis including a PRISMA flow diagram (Page et al 2021) for details of the flow of study selection, a table of characteristics for included studies, risk of bias assessment, forest plots for each outcome with accompanying descriptive text.

[PRISMA diagram]

Description of included studies

We identified 61 eligible studies. The study characteristics can be found in Table 1. Data from studies contributed with at least one outcome with quantitative data (total of 9745 participants), which included adults from multiple countries. The mean age of participants was 42.4 years (range 15 to 72 years), with a mean proportion of 0.58 female participants (range 0 to 0.86). Included studies allocated the participants to treatment lasting between 4 to 12.9 weeks (median, 6 weeks).

Primary outcome: reduction in anhedonia scores at 8 weeks (from 4 to 12 weeks)

Figure X Forest plot for symptoms of anhedonia (primary outcome) comparing pro-dopaminergic interventions vs placebo for individuals with anhedonia at 4-12 weeks (primary timepoint). SMD: standardised mean difference, 95%CI: 95% confidence intervals, SD: standard deviation.

6 studies contributed with data to the meta-analysis with a total of 2079 participants (1143 allocated to pro-dopaminergic interventions, 936 allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of -0.244 (95%CI from -0.456 to -0.032). The study results presented an I2 value of 67.8% (95%CI from 23.6% to 86.4%) and a tau2 value of 0.025 (accounting for a 95% prediction interval from -0.74 to 0.252).

Risk of bias

Sensitivity analyses

Sub-group and meta-regression analyses

We did not find sufficient data (i.e. at least ten studies) to conduct a meta-regression for the investigated predictors. Insufficient data existed for age, proportion of females, anhedonia baseline score, anxiety baseline score, reward baseline score, and treatment duration.

Secondary outcome: Reduction in mean anxiety score at 8 weeks (from 4 to 12 weeks)

Figure X Forest plot for symptoms of anxiety (secondary outcome) comparing pro-dopaminergic interventions vs placebo for individuals with anxiety at 4-12 weeks (primary timepoint). SMD: standardised mean difference, 95%CI: 95% confidence intervals, SD: standard deviation.

11 studies contributed with data to the meta-analysis with a total of 3517 participants (2077 allocated to pro-dopaminergic interventions, 1440 allocated to pill placebo.

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of -0.166 (95%CI from -0.243 to -0.088). The study results presented an I2 value of 0% (95%CI from 0% to 60.2%) and a tau2 value of 0 (accounting for a 95% prediction interval from -0.251 to -0.08).

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age, proportion of females, and anxiety baseline score. Insufficient data existed for anhedonia baseline score, reward baseline score, and treatment duration.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was not significant.
  • The meta-regression for anxiety baseline score was not significant.

Secondary outcome: Dropouts due to any reason

Figure X Forest plot for dropouts due to any reason for the comparison of pro-dopaminergic interventions vs placebo at 4-12 weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

50 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of -0.092 (95%CI from -0.315 to 0.13). The study results presented an I2 value of 67.6% (95%CI from 56.6% to 75.8%) and a tau2 value of 0.334 (accounting for a 95% prediction interval from -1.274 to 1.089).

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age, proportion of females, and anxiety baseline score. Insufficient data existed for anhedonia baseline score, reward baseline score, and treatment duration.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was not significant.
  • The meta-regression for anxiety baseline score was not significant.

Secondary outcome: dropouts due to side effects

Figure X Forest plot for dropouts due to adverse events for the comparison of pro-dopaminergic interventions vs placebo at 4-12 weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

40 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of 0.552 (95%CI from 0.227 to 0.876). The study results presented an I2 value of 45.7% (95%CI from 21.1% to 62.6%) and a tau2 value of 0.395 (accounting for a 95% prediction interval from -1.274 to 1.863).

Meta-regressions were conducted for age, sex, and baseline anxiety score. Of these, only sex was a significant predictor of dropouts due to adverse events with a beta=-2.960, SE=1.21 95%CI (-5.423; -0.497) and an R^2=40.63% and p=.022.

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age, proportion of females, and anxiety baseline score. Insufficient data existed for anhedonia baseline score, reward baseline score, and treatment duration.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was significant (beta = -2.96, 95% CI from -5.423 to -0.497).
  • The meta-regression for anxiety baseline score was not significant.

Secondary outcome: nausea

Figure X Forest plot for nausea for the comparison of pro-dopaminergic interventions vs placebo at 4-12 weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

24 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of 0.378 (95%CI from 0.169 to 0.586). The study results presented an I2 value of 27.4% (95%CI from 0% to 56%) and a tau2 value of 0.065 (accounting for a 95% prediction interval from -0.19 to 0.945).

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age and proportion of females. Insufficient data existed for anhedonia baseline score, anxiety baseline score, reward baseline score, and treatment duration.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was not significant.

Secondary outcome: headache

Figure X Forest plot for headaches for the comparison of pro-dopaminergic interventions vs placebo at 4-12 weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

26 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of 0.149 (95%CI from 0.029 to 0.269). The study results presented an I2 value of 0% (95%CI from 0% to 43.2%) and a tau2 value of 0 (accounting for a 95% prediction interval from 0.008 to 0.291).

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age and proportion of females. Insufficient data existed for anhedonia baseline score, anxiety baseline score, reward baseline score, and treatment duration.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was not significant.

Secondary outcome: insomnia

Figure X Forest plot for insomnia for the comparison of pro-dopaminergic interventions vs placebo at 4-12 weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

22 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of 0.583 (95%CI from 0.354 to 0.812). The study results presented an I2 value of 5.4% (95%CI from 0% to 37.7%) and a tau2 value of 0.03 (accounting for a 95% prediction interval from 0.15 to 1.017).

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age and proportion of females. Insufficient data existed for anhedonia baseline score, anxiety baseline score, reward baseline score, and treatment duration.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was not significant.

Secondary outcome: constipation

Figure X Forest plot for dropouts due to adverse events for the comparison of pro-dopaminergic interventions vs placebo at 8 (4-12) weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

19 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of 0.439 (95%CI from 0.222 to 0.656). The study results presented an I2 value of 0% (95%CI from 0% to 48.9%) and a tau2 value of 0 (accounting for a 95% prediction interval from 0.183 to 0.696).

There were no significant subgroup differences between participants assigned to different classes of drug.

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age, proportion of females, and treatment duration. Insufficient data existed for anhedonia baseline score, anxiety baseline score, and reward baseline score.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was not significant.
  • The meta-regression for treatment duration was not significant.

Secondary outcome: dizziness

Figure X Forest plot for dizziness for the comparison of pro-dopaminergic interventions vs placebo at 8 (4-12) weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

22 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of 0.543 (95%CI from 0.28 to 0.805). The study results presented an I2 value of 26.5% (95%CI from 0% to 56.4%) and a tau2 value of 0.084 (accounting for a 95% prediction interval from -0.115 to 1.2).

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age and proportion of females. Insufficient data existed for anhedonia baseline score, anxiety baseline score, reward baseline score, and treatment duration.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was not significant.

Secondary outcome: dry mouth

Figure X Forest plot for dry mouth for the comparison of pro-dopaminergic interventions vs placebo at 4-12 weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

24 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed an effect favouring pro-dopaminergic interventions with a SMD of 0.751 (95%CI from 0.519 to 0.983). The study results presented an I2 value of 36.1% (95%CI from 0% to 61%) and a tau2 value of 0.08 (accounting for a 95% prediction interval from 0.126 to 1.376).

Sub-group and meta-regression analyses

We found enough data (i.e. at least ten studies) to conduct a meta-regression for age, proportion of females, and treatment duration. Insufficient data existed for anhedonia baseline score, anxiety baseline score, and reward baseline score.

  • The meta-regression for age was not significant.
  • The meta-regression for proportion of females was significant (beta = -2.071, 95% CI from -3.834 to -0.308).
  • The meta-regression for treatment duration was not significant.

Secondary outcome: vomiting

Figure X Forest plot for vomiting for the comparison of pro-dopaminergic interventions vs placebo at 4-12 weeks. OR: odds ratio, 95%CI: 95% confidence intervals.

5 studies contributed with data to the meta-analysis with a total of 0 participants ( allocated to pro-dopaminergic interventions, allocated to pill placebo).

The comparative effect of pro-dopaminergic interventions versus placebo showed a comparable effect not excluding the null effect with a SMD of 0.641 (95%CI from -0.104 to 1.386). The study results presented an I2 value of 2.2% (95%CI from 0% to 79.7%) and a tau2 value of 0 (accounting for a 95% prediction interval from -0.204 to 1.485).

Sub-group and meta-regression analyses

We did not find sufficient data (i.e. at least ten studies) to conduct a meta-regression for the investigated predictors. Insufficient data existed for age, proportion of females, anhedonia baseline score, anxiety baseline score, reward baseline score, and treatment duration.

Reporting bias

Summary of evidence tables

Abbreviations

References